My biggest gripe about custom maps is that many of them are just too big. As in, it takes more than 60 seconds to get from one end of them to the other. A ton of stuff can happen in 60 seconds. A flag can be returned, a player can die and re-spawn at least 4 times, and a sentry gun can be built in taken all the way to level 3 in this span of time. Anyone can tell you running for 5 minutes just to get killed in 5 seconds is just no fun. Therefore, defining the pacing to the level is pivotal.

A great thing about working in 3D is that changing the scale of most objects is easy and lossless. However, when Displacement Maps are scaled, it can come out a bit pointier and different than was originally envisioned. Also, scaling an object does not scale its texture map. With this in mind, I set out to design the pacing of Staredown by “sketching” in with basic geometry and textures that only give the impression of what is to come.

To find a starting point, I ran through various levels by myself, comparing how they function without any players to my memories of playing with a full server. The flag room in 2Fort is a great size, it easily holds 3 difficult to kill sentry guns, but even 1 well placed gun can slow down enemy offense long enough to must a defending force. For the offense it offers 2 angles of attack and is can be crossed quickly by scout and spy classes. With this in mind, this Mapper’s Reference shows some very important numbers and rules including 2Fort Room sizes. Staredown’s Flag room is just slightly taller and longer, but not significantly so. The base scale is set.

From this room, I can now build up the rest of the level based on the sketch. This base is meant to be on the interior of a mountain, but I’ll be using a displacement map for that so no roof for now… I’ll just have to use the power of imagination or something. That won’t works so well for a floor, so for now a flat concrete will keep players from falling ad infinitum. The rest of the walls come together fairly easily and I only spend extra time on some of the doorway carvings. After a few test and tweaks, I worry about the front and first wall being a little long, but overall it is satisfactory.

Now to make the outdoors based on the sketch. I want a fairly smooth doughnut shape so my first thought is Cube - Cylinder = Awesome. WRONG! Carving with cylinders is a slow slow bad idea. My system is no slouch either! New approach: The Arch brush can make a complete circle without a center, so I did that, then connected a hallowed out Cylinder to it. Delete the roof of the cylinder, and there we have our doughnut shape, ready for displacement mapping if we want, and at a fraction of the cycles. Excited, I threw the base onto my terrain and compiled the map. Before me was a vast wasteland, and the thing I hated most… a 90 second run to the opposite base.

         

The Fix
It turns out that the current size of the map is roughly 3 bases in diameter! After burning my initial sketch, I remembered that I can just scale down the terrain that I made. To determine the scale each circuit around the doughnut was timed. Every time the map got smaller I was a bit saddened, because really wanted a nice arena in the center. Soon, I figured, it wouldn’t even be worth going down there. It is necessary, in these situations, to believe in the original design goal and not get attached to secondary ideas. In a radical swoop, I cut down the scale to the bare minimum needed to contain the base. The arena was reduced to a semi-circle, but now it only takes about 28 seconds to get to the enemy base.

In a similar move, the 3 stage ramp from the bottom of the arena to base level was cut to two because even as fast as 28 seconds is, it’s not fast enough. After a 16 second spawn time, that only leaves 15 seconds to actually find, attack, and secure a dropped flag. The arena route needs to be as fast as possible so the slopes are as deep as possible. Once again a few more scaling tweaks were needed on the ramps to them just right, but by the end, players can get down and up in about 16 seconds. Players willing to take a small amount of falling damage can probably get there even faster.
So right now the world seems a little too big still, but the timing is about right. Once a few props are in, it should seem much smaller and faster. Fortunately I wasted no time putting down and tweaking props that would only end up getting moved once I found that it took too long to get from end to end. The textures used have just enough contrast to give the impression of what I’m going for. The floor plan is ready, lets see if we can fill it with stuff that makes it fun
Next up: Where am I now?

This entry was posted on Friday, December 14th, 2007 at 7:04 pm and is filed under Level Design, Team Fortress 2, Development. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.